Localisation and web accessibility: a comparative study of the transfer of accessibility information through CAT tools

(1)

Master

Reference

Localisation and web accessibility: a comparative study of the transfer of accessibility information through CAT tools

PACATI, Isotta

Abstract

By conducting a comparative study of two CAT tools, SDL Trados Studio 2017 and MemoQ v8.7, this research project aimed to examine the impact of these tools on the achievement of accessibility conformance during the web localisation process. Two different approaches were adopted to investigate this matter. The first stage comprised a descriptive approach, in which the researcher analysed the CAT tools' performance and determined whether they could support and transfer a selection of qualities related to accessibility embedded in an HTML5 code. The results indicated that the two systems supported and transferred the majority of these elements, but not all of them. The second stage of the research consisted of a tool evaluation carried out by ten novice web localisers. The goal was to measure the functional suitability of the CAT tools and analyse the influence of participants' knowledge of accessibility on the final target product.

PACATI, Isotta. Localisation and web accessibility: a comparative study of the transfer of accessibility information through CAT tools. Master : Univ. Genève, 2020

Available at:

http://archive-ouverte.unige.ch/unige:131301

Disclaimer: layout of this document may differ from the published version.

(2)

Isotta Pacati

Localisation and web accessibility: a comparative study of the transfer of

accessibility information through CAT tools

Directrice : Silvia Rodriguez Va ́zquez Jurée : Luci ́a Morado Vázquez

Mémoire présenté à la Faculté́ de traduction et d’interprétation (Département Traitement Informatique Multilingue), pour l’obtention de la Maîtrise universitaire en traduction, mention Technologies de la traduction

Université́ de Genève Année Académique 2019-2020

Janvier 2020

(3)

(4)

Déclaration attestant le caractère original du travail effectue ́

J’affirme avoir pris connaissance des documents d’information et de prévention du plagiat émis par l’Université́ de Genève et la Faculté́ de traduction et d’interprétation (notamment la Directive en matière de plagiat des étudiant-e-s, le Règlement d’études des Maîtrises universitaires en traduction et du Certificat complémentaire en traduction de la Faculté́ de traduction et d’interprétation ainsi que l’Aide-mémoire à l’intention des étudiants préparant un mémoire de Ma en traduction).

J’atteste que ce travail est le fruit d’un travail personnel et a été́ rédigé de manière autonome.

Je déclare que toutes les sources d’information utilisées sont citées de manière complète et précise, y compris les sources sur Internet.

Je suis conscient-e que le fait de ne pas citer une source ou de ne pas la citer correctement est constitutif de plagiat et que le plagiat est considéré comme une faute grave au sein de l’Université, passible de sanctions.

Au vu de ce qui précède, je déclare sur l’honneur que le présent travail est original.

Nom et prénom :

Lieu / date / signature :

(5)

Acknowledgments

I would like to express my very great appreciation to my supervisor Silvia Rodriguez Vázquez, who provided many interesting ideas at the beginning of this journey and helped me develop this thesis idea. Although I was lacking, thank you for always advising and guiding me until the very end. Thank you for your precious feedback and continuous support throughout the whole process. I would also like to thank Lucìa Morado Vázquez, who introduced me to the world of localisation and have sparked my interest in this field with her passion and enthusiasm.

Un grazie infinito alle mie compagne di Facoltà Flaminia, Sofia, Silvia, Paola, Raffaella e Benedetta, con le quali ho condiviso gli ultimi cinque anni. Grazie per essere state al mio fianco, per le ore passate chine sui libri, per le serate passate a ridere e per avermi aiutato e incoraggiato durante i momenti difficili.

Un ringraziamento speciale va alle mie amiche di una vita Alice S., Matilde e Alice M., che da sempre mi hanno incoraggiato e non mi hanno mai fatta sentire sola, nonostante la distanza.

Grazie per le continue parole di supporto e per essere sempre con me, ovunque vada.

Ai miei genitori, a cui devo tutto. Grazie per tutti i vostri insegnamenti e per aver sempre creduto in me.

(6)

Abstract

In recent years, multilingual accessibility has become an issue of increasing relevance. By conducting a comparative study of two CAT tools, SDL Trados Studio 2017 and MemoQ v8.7, this research project aimed to examine the impact of these tools on the achievement of accessibility conformance during the web localisation process.

Two different approaches were adopted to investigate this matter. The first stage comprised a descriptive approach, namely a tool descriptive analysis, in which the researcher analysed the CAT tools’ performance and determined whether they could support and transfer a selection of qualities related to accessibility embedded in an HTML5 code. The results indicated that the two systems supported and transferred the majority of these elements, but not all of them. In addition, the second stage of the research consisted of a tool evaluation carried out by ten novice web localisers. The goal was to measure the functional suitability of the CAT tools and analyse the influence of participants’ knowledge of accessibility on the final target product. In this case, the findings suggested that CAT tools may have a positive impact on the achievement of localisation when localisers are not accessibility-savvy, as they offer useful additional information about the feature’s context, and proved that localisers who know how to implement accessibility best practices can improve the overall accessibility conformance of the final product.

Keywords: web localisation, web accessibility, HTML, CAT tools, EAGLES, CAT tool evaluation

(7)

1 Introduction

This master’s thesis focuses on the process of web localisation, more specifically on the localisation of accessibility information: it investigates whether Computer-Aided Translation (CAT) tools, used for the localisation of HTML files, can support and transfer all the features related to accessibility embedded in the code, and help localisers produce an accessible target file.

1.1 Research context

People's way of communicating and interacting has changed significantly since the advent of the Internet and the World Wide Web, together with the development of other digital technologies. In recent decades, information and communication technologies (ICT) have continued to grow exponentially (United Nations 2006, 169). According to the International Telecommunication Union (ITU), in 2018 more than half of the world population, around 51%

(ITU 2018, 3) used the Internet, and it is expected to reach 75% by 2025 (ibid., 13). These figures highlight the central role that ICTs, principally the Web, have in our daily lives. As digital technologies affect nearly every aspect of life, every Internet user should access freely to the information provided online. However, this is not always the case, as shown in the Disability and Development Report, published by the United Nations' Department of Economic and Social Affairs: the percentage of people with disabilities which uses the Internet (19%) is lower compared to the percentage of people without disabilities (36%), as of 2011 (United Nations 2019, 173). Several reasons influence the lower use of the Internet, including accessibility-related issues. In other words, people with disabilities may face difficulties in accessing the information included in webpages if proper accessibility practices are not implemented.

ICTs can be a double-edged sword for disabled people. On the one hand, they "represent a powerful opportunity to improve quality of life, enhance inclusion and social engagement and make independent living possible" (ibid., 169), as users can have access to various online services, including public services, and e-learning materials, text-to-voice devices, etc.

Therefore, digital technologies can help people with disabilities engage in society (ibid.). On the other hand, if digital technologies are not accessible, the opportunity of developing and maintaining their social life is taken away from them. Hence, we can affirm that "[u]nequal access to digital technologies brings about unequal participation in society" (Van Dijk 2005,

(17)

15). Moreover, unequal access creates a ‘digital divide’ (Goggin 2018, 1) between people with disabilities and people without disabilities.

1.1.1 Web accessibility

The general term ‘accessibility’ refers to the “quality of being easily reached, entered, or used by people who have a disability”¹. Consequently, accessibility in the context of information and communication systems, in particular of websites, defines the extent by which the webpage is considered accessible, meaning barrier-free, to everyone: every user, including people with disabilities and the elderly, should have access to all the information available on the Internet (Yates 2005 in Qadri & Banday 2009, 2). In this specific case, then, it would be more accurate to talk about ‘web accessibility’, rather than just ‘accessibility’.

The Web has facilitated access to information but, at the same time, it has aggravated the problem of exclusion of people with disabilities. This issue has pushed organisations, such as the World Wide Web Consortium (W3C), and governments to undertake measures that would ensure web accessibility, which resulted in several initiatives and enactment of legislations and laws (Qadri & Banday 2009, 3-4).

According to W3C, web accessibility “means that websites, tools, and technologies are designed and developed so that people with disabilities can use them” (W3C/WAI 2019), and it can be included in the concept of ‘universal design’. The latter refers to the creation and development of products that can be used by everyone, within the widest range of situations (Henry, Abou-Zahra, & Brewer 2014, 1), including websites.

To propose a unified definition, Petrie, Savva and Power (2015) analysed 50 definitions of web accessibility. Their work resulted in the following definition:

“all people, particularly disabled and older people, can use websites in a range of contexts of use, including mainstream and assistive technologies; to achieve this, websites need to be designed and developed to support usability across these contexts” (Petrie et al. 2015, 1)

1 https://www.lexico.com/en/definition/accessibility [11.06.2019]

(18)

According to the authors, having a unified definition that includes all the six core concepts² found in the other definitions avoids running the risk of creating a product, in our case a website, which is not accessible (ibid.).

As the Web is becoming more and more complex, web developers and web designers should know how to make their webpages accessible. To do so, in 1997, W3C launched the Web Accessibility Initiative (WAI), which “provides an international forum for collaboration between industry, disability organizations, accessibility researchers, government, and other interested in Web accessibility” (W3C/WAI 2019a). Among all its tasks, WAI develops a set of guidelines used by governments and organisations as international standards for web accessibility (Qadri & Banday 2009, 2): the Web Content Accessibility Guidelines (WCAG) (Kirkpatrick et al. 2018), which we will illustrate in Chapter 2.

1.1.2 Disability

The concept of disability is complicated and multidimensional, and many attempts have been made to define it from numerous perspectives (Altman 2001, 97-98). As stated in the United Nations’ Convention on the Rights of Persons with Disabilities (CRPD), people with disabilities “include those who have long-term physical, mental, intellectual or sensory impairments which in interaction with various barriers may hinder their full and effective participation in society on an equal basis with others” (United Nations 2006, 4). This definition introduces two important concepts for our research: impairment and barrier.

The social model of disability redefined the concept by making a distinction between

‘disability’ and ‘impairment’: “[d]isability is something imposed on top of our impairments, by the way we are unnecessarily isolated and excluded from full participation in society.” (UPIAS 1975 in Shakespeare, 2010, 198). Therefore, we can state that impairment is a term that refers to a physical or mental limitation, while disability refers to the social exclusion (ibid.): by providing accessibility solutions, the individual has an impairment but can still take part in social activities. This argument also applies to web accessibility. For instance, if a blind person navigates a website in which the right accessibility practices are implemented, said person will still be visually impaired but will be able to access the information.

Moreover, the Disabled People’s International (DPI) included the concept of ‘barrier’ in its definition of disability, which is described as “the loss or limitation of opportunities to take part

2 Six core concepts: groups of users, characteristics, needs of users; what users should be able to do; technologies used; characteristics of the website; design and development of the website; characteristics of the situations of use.

(19)

in the normal life of the community on an equal level with others due to physical and social barriers” (Oliver 1996, 33). In the context of web accessibility, we can talk about digital barriers that prevent people with disabilities from accessing the information, namely obstacles that can be avoided if sufficient practices are implemented. Among the numerous barriers that exist in the digital world, we find audio content without captions or transcript that are not accessible to people with auditory disabilities, complex navigation mechanisms and layout that can be confusing for people with cognitive, learning and neurological disabilities, websites that do not provide keyboard support for people with physical impairments, and non-text content that does not have an equivalent text alternative for people with visual impairments (W3C/WAI 2017).

The difficulty in defining the concept of disability is also due to the complexity of its nature. It is fundamental to recognise that disability is an evolving concept and to recognise the diversity of people with disabilities (United Nations 2006, 1-2). There is a variety of impairments that need to be considered when elaborating standards and practices related to accessibility. For instance, visual, auditory, and motor disabilities, together with cognitive problems, were targeted by WAI when drafting the first version of the WCAG (Chisholm, Vanderheiden &

Jacobs 1999).

1.1.3 Assistive technologies

As we have introduced in the previous section, physical limitations and mental impairments can prevent people with disabilities to use digital products, in our case websites, which are not designed according to their needs. However, there are many ways for disabled people to navigate the Web, depending on their abilities (W3C/WAI 2017a). One option to access the information online is to provide the individual with a tool, such as assistive technology (AT) (Vanderheiden 1998, 30).

In general, AT refers to “any item, piece of equipment, or product, whether it is acquired commercially, modified, or customized, that is used to increase, maintain, or improve the functional capabilities of individuals with disabilities” (WHO 2011, 101). ATs have the advantage to be adapted to the needs and abilities of the individual, in addition to increasing independence and participation (Vanderheiden 1998; WHO 2011). In this research, we are interested in ATs that enable individuals with disabilities to use the Web.

The authors of WCAG defined AT as a “hardware and/or software that acts as a user agent, or along with a mainstream user agent, to provide functionality to meet the requirements of users with disabilities that go beyond those offered by mainstream user agents” (Kirkpatrick et al.

(20)

2018), which provides functionalities such as alternative presentations, alternative input methods, additional navigation or orientation mechanisms, and content transformation. Useful ATs to access a graphical user interface (GUI) are visual reading assistants, such as screen magnifiers, that allow the user to change text font, size, spacing, and other components to improve the visual readability; screen readers, which allow blind people to access information for non-text content; text-to-speech software; alternative keyboards, for instance, head pointers or single switches, and alternative pointing devices, which simulate respectively the keyboard and the mouse pointing and button activations (ibid.).

1.2 Motivation

As stated at the beginning of this thesis, the Internet has become a channel to convey information about numerous topics, including health and education, and has become a place where people can have access to many public services. “Accessing general information online enables people with disabilities to overcome any potential physical, communication and transport barriers in accessing other sources of information” (WHO 2011, 183-184); therefore, ICT equipment and services should be designed to benefit not only the wider population but also people with impairments (ibid.). However, only in Europe, less than 10% of websites are accessible, and 5% of the European population does not use the Internet due to an impairment (European Commission 2019, 1).

A multi-sectoral and multi-stakeholder approach can be an option to guarantee accessibility:

“[g]overnments, industry and end-users all have a role in increasing accessibility” (WHO 2011, 186). As suggested by WHO, the industry itself plays an important role in ensuring access to accessible information. It is fundamental for all actors involved in the web cycle to gain a certain level of expertise to guarantee a sufficient degree of accessibility (Rodríguez Vázquez 2013, 384). Among these actors, we can also include web localisers. Localisation, as we will discuss in greater detail in Chapter 2, can be described as the process of adapting a digital product to a language and culture. With the term web localisation, we define a specific type of localisation that consists in the adaptation of the content included in a website (see Section 2.1.1).

A study aiming to define localisers’ role during accessibility assessment tasks showed that localisers’ participation would contribute to the achievement of web accessibility, in particular for text and graphic content in the context of multilingual websites (ibid, 387). There are many

(21)

aspects that a localiser should consider, as highlighted by Torres del Rey and Morado Vázquez (2019):

“If content is […] linguistically, culturally, semantically, pragmatically and technically accessible for source locale users, those synergies must be reconstructed in a different locale, with different expectations and experiences among communities of people with disabilities. It is therefore logical to assume that success in localizing accessible content depends on how that content is transferred and, if necessary, adapted and transformed linguistically, culturally and technically in relation to its surrounding context, and linguistic, cultural, semantic, pragmatic and technical features, and according to target users and use environment.” (Torres del Rey & Morado Vázquez 2019, 6)

In addition to the above, we also need to consider the influence that the tools used during the web localisation process, such as CAT tools, can have on the achievement of web accessibility.

Do they support the localiser’s effort in ensuring accessibility? Do they have a positive or negative impact? With these questions in mind, through our work, we aim at contributing to the current research in the field by analysing the impact that CAT tools have on the degree of accessibility achieved in the final target product in the context of localisation.

1.3 Research questions and hypotheses

As introduced at the beginning of this master’s thesis, this work will focus on one of the steps (use of CAT tools) of the process of web localisation (defined in Chapter 2). In the previous sections, we reviewed how people with disabilities use the Web and highlighted the importance of web accessibility. In this section, we outline the main objective of this research, the research questions and the hypotheses on which we based the design of our study.

Web accessibility does not only concern the design and creation of a web-based GUI, but it should also be taken into account during the localisation process. One could assume that accessible source files, if localised correctly, will produce accessible target files. On the contrary, if the information and techniques supporting accessibility are ignored or not transferred correctly, the degree of accessibility of the target file will be inferior in comparison to the source, and, consequently, people with disabilities being part of the target audience will not be able to access the information properly. In the particular context of this thesis, we are interested in what we call ‘accessibility features’. By this term, we understand the

(22)

characteristics that help achieve accessibility, namely elements (coding elements, attributes, text units, etc.) that can be embedded in the code to ensure accessibility.

Computer-Aided Translation (CAT) tools are often employed by localisers to translate HTML- based files. These tools usually support this type of files: they identify and isolate the translatable or localisable information and protect the code that in principle should not be modified. As certain accessibility features are often embedded in the code, there is the risk that they might be overlooked or even not supported by the CAT tool in use. Taking into account this initial premise, our research aims at exploring the following questions:

RQ1. Can CAT tools support and therefore transfer all relevant accessibility features when processing an HTML5 file?

And therefore,

RQ2. Is the resulting target file accessible?

The relevant literature (Chapter 2), more specifically researches related to multilingual web accessibility and localisation, together with researches on HTML-format file support in CAT tools, led us to formulate the following hypothesis:

H1. CAT tools do not support and transfer all the accessibility features.

To answer the second question and examine the functional correctness of the tools studied, we conducted an experiment, which we will describe in Chapter 3. The experiment sought to confirm or reject the following hypotheses:

H2. The functional completeness of the tool used has an impact on the final degree of the accessibility achieved.

H3. The functional appropriateness of the tool used has an impact on the final degree of accessibility achieved.

H4. The participants’ level of knowledge of accessibility has an impact on the final degree of accessibility achieved.

1.4 Methods

To answer the research questions, we adopted two different approaches. To determine if CAT tools can support and transfer the information features concerning accessibility, we adopted a descriptive approach and performed a tool descriptive analysis of two CAT tools: SDL Trados Studio 2017 and MemoQ v8.7. The researcher examined the tools’ features and functionalities

(23)

when processing an HTML5 file to determine if the accessibility features selected for our study (defined in Chapter 3) were supported correctly and had the potential to be transferred to the target document accordingly.

As we introduced in the previous section, to answer the second question we conducted an experiment with novice web localisers. Firstly, participants were asked to answer a preliminary questionnaire to determine their background, their level of knowledge of accessibility and localisation, and their experience with CAT tools. This questionnaire helped define the hypotheses and the test that followed. Secondly, participants were asked to localise an HTML5 file with both CAT tools and answer a post-evaluation questionnaire. The design of the study was based on the seven steps proposed by the Expert Advisory Group on Language Engineering Standards (EAGLES) (EAGLES 1999), which will be explained in Chapter 3.

1.5 Structure of the thesis

In this introductory chapter, we illustrated the research context, focusing on the definition of disability and the importance of web accessibility, the questions and hypotheses, and two methods used in this thesis. The following chapter (Chapter 2) provides an overview of the fundamental concepts of our research, namely localisation, accessibility, and HTML, related works, and a comparison of theories. Chapter 3 includes the methodology established to carry out both the tool descriptive analysis and the user evaluation. Chapter 4 describes the data collected through the first stage of our research, while Chapter 5 illustrates the results of the experiment carried out by participants and discusses the findings of our work. Finally, in Chapter 6, we draw the conclusions of our work and set new avenues for future research in this area.

(24)

2 Literature Review

In this chapter, we analyse the fundamental concepts of this master’s thesis: localisation, accessibility, and CAT tools. In the first section (2.1), we illustrate how the industry and Translation Studies (TS) scholars define localisation, and describe in more details the specific type of localisation which will be the object of our research, namely website localisation, as well as the most popular markup language used to develop web content, HTML. Section 2.2 deals with web accessibility, specifically the set of guidelines and techniques that are usually employed to make a website accessible. In Section 2.3, we examine the relationship between the topics reviewed in the two previous sections, we define the concept of ‘accessibility features’ and provide an overview of the most relevant accessibility success criteria for web localisation. Finally, in Section 2.4, we review the main functionalities of Computer-Aided Translation (CAT) tools, prior studies on CAT tools, localisation and accessibility, and the evaluation methods that exist to assess this type of software.

2.1 Localisation

In this section, we introduce the first fundamental notion of this thesis: localisation. Before analysing how Translation Studies (TS) scholars and the industry define this concept, we will briefly retrace its history: how it all started and how it developed through the decades. People in the industry started talking about localisation in the late 1970s and early 1980s when personal computing and software became popular among people who did not possess programming skills. Economic reasons were central for the evolution of localisation, as companies such as Microsoft and Oracle wanted to export their products, already popular in the U.S., in other countries like Japan and the so-called FIGS countries (France, Italy, Germany, and Spain) (Jiménez-Crespo 2013, 8).

At first, developers hired linguists to translate textual strings and establish translation practices.

However, they soon realised that separating the software from the translation only posed technical challenges, as it required translators to possess a basic knowledge of programming (ibid., 9). This factor and the formation and development of multi-language vendors (MLVs) in the mid-1980s, together with an extended outsourcing model which became popular in the 1990s, are among the factors that started the transition from translation to localisation (Esselink 2000, 5-6). Nevertheless, the constant demand for new target languages led to a reshaping in the industry in the 2000s. At present, the localisation industry has become more and more

(25)

complex, since new technological developments have been introduced (Jiménez-Crespo 2013, 9).

But how do experts in the industry and TS scholars define this concept? The term comes from the notion of ‘locale’, defined by the ISO standard 17100 (ISO 2015) as the “set of characteristics, information or convention specific to the linguistic, cultural, technical, and geographical convention of a target audience” (ibid.). Likewise, Pym (2004) defines it as the ensemble of a particular variety of a language and local convention (such as currency, dates, etc.) (ibid., 2).

Localisation can be defined by maintaining this notion of ‘locale’, as we can see in the definition given by the now-defunct Localisation Industry Standard Association (LISA): “[l]ocalization involves taking a product and making it linguistically and culturally appropriate to the target locale (country/region and language) where it will be used and sold” (LISA 2003, 13 in Jiménez-Crespo 2013, 13). However, in 2007, LISA itself expanded its own definition to include a large range of service and adopted the notion of ‘market’ instead of ‘locale’:

“[l]ocalization is the process of modifying products or services to account for differences in distinct markets” (ibid.). This second definition highlights the relevance of economic reasons that lie behind the localisation industry, as it is part of a broader and global cycle.

Similarly, the Globalisation and Localisation Association (GALA) also adopted the notion of

‘locale’ in its definition and described localisation as “the process of adapting a product or content to a specific locale or market” (GALA 2015). In addition, GALA listed all the elements that are included, besides translation, in the localisation process: “adapting graphics to target markets, modifying content to suit the tastes and consumption habits of other markets, adapting design and layout to fit translated text, converting to local requirements (such as currencies and units of measure), using proper local formats for dates, addresses, and phone numbers, and addressing local regulations and legal requirements” (ibid.). Through this definition, we realise the complexity of the whole process: localisation does not consist only of the adaptation of the original text, but there are numerous aspects that localisers should consider. Among these, we could also add what we define as ‘accessibility features’, that we will further explain later in this chapter.

From a TS perspective, we can identify two major trends that are followed by scholars. On one hand, localisation is considered as a ‘translation-related phenomena’, as a mere translation modality shaped by technological and project-based features. On the other hand, the second trend focuses on the description of industrial practices through a professional approach

(26)

(Jiménez-Crespo 2013, 17). For instance, concerning the latter trend, Dunne (2006) gave the following definition, which includes the concept of locale and adds the notion of non-text content:

"[t]he processes by which digital content and products developed in one locale (defined in terms of geographical area, language and culture) are adapted for sale and use in another locale. Localisation involves: (a) translation of textual content into the language and textual conventions of the target language, (b) adaptation of non-textual content (from colors, icons and bitmaps, to packaging, form factors, etc.) as well as input, output and delivery mechanism to take into account the cultural, technical and regulatory requirements of that locale (ibid., 4)."

As we previously mentioned, to understand the concept of localisation, we need to consider it as part of a broader and global cycle, “a much wider complex of interrelated processes known as GILT” (Jiménez-Crespo 2013, 24): Globalisation – Internationalisation – Localisation and Translation cycle.

In the context of the language industry, with the term globalisation, we refer to the broader processes in the cycle necessary to offer products and activities on a global scale. This concept not only includes language-related aspects such as multilingual communication but also practices related to commerce and trade policies (GALA 2015a).

Likewise, internationalisation refers to the stage that precedes the development of a digital product (Jiménez-Crespo 2013, 25). This process can facilitate the localisation task, as it ensures that a product “can be adapted to various languages and regions without requiring changes to the source code” (GALA 2015b). In the context of our research, the internationalisation of a website allows users from different locales to access equally to information. This process is usually carried out during the product development cycle prior to the localisation process (Esselink 2000, 2-3). Since web localisation is at the centre of website internationalisation, we refer to the publication of a multilingual website as ‘website globalisation’, which includes the internationalisation of the back-end software, the design of a multilingual architecture, and the localisation of the site’s content (ibid., 4-6). In this research, we will focus on the latter to illustrate how localisation practitioners can obtain a multilingual website.

The following step concerns localisation, which we have just defined above. In the GILT cycle, this stage includes preparation, management, engineering and quality assurance (QA).

Therefore, the fourth step, the translation process, can be described as the transfer of textual

(27)

material by the translator-localiser (Jiménez-Crespo 2013, 26). In this research, we will focus on the third and fourth stages, namely the actual localisation and translation processes, in which the product is adapted to the target locale.

Last, it is worth mentioning here that localisation, as a general term, refers to numerous types of localisation practices, such as software, videogame, application and website localisation. In this thesis, we will focus on the latter, which we will expand on in the following section.

2.1.1 Web localisation

Website localisation, or just web localisation, emerged in the latest 1990s and it is often included in more general definitions of localisation. For instance, Esselink (2000) describes the general notion as the “translation and adaptation of a software or a web product” (ibid., 1).

Therefore, we can state that web localisation is the process of adapting the content of a website to “make it accessible, usable, and culturally suitable to a target audience” (Sandrini 2008, 9).

In this specific case, the author highlights how the content should be adapted to a specific audience, in all its diversity. As we will see in Section 2.2 and later in Chapter 3, the audience comprises diversified individuals with varying abilities and, consequently, the product should be adapted according to everyone’s needs, including people with disabilities’.

Moreover, in the attempt of defining localisation in a more general way, Jiménez-Crespo (2013) described web localisation as “a process by which interactive digital texts are modified to be used in different linguistic and sociocultural contexts, guided by expectations of the target audience and the specifications and degree requested by initiators” (ibid., 20). Here, specifications and the degree requested by initiators could also refer to the information concerning accessibility that should be respected and transferred correctly to obtain an accessible product in the target locale.

Before progressing further with our literature review, it is fundamental to understand the type of content that the localiser is normally asked to adapt and translate, as localisation does not only concern language issues (Pym 2011, 5). Sandrini (2008) divides the website content, composed of digital assets, in six categories: common content, multimedia assets, applications- bound assets, transactional assets, and community assets (ibid.). The localisable content can be included in those six categories and concerns mainly text units, which can be found in the title, descriptions, keywords, menus, hyperlinks, description of non-text content (such as images, videos, graphics, etc.), and in audio-visual files (Pym 2011, 2).

(28)

Website content is subject to a ‘content life cycle’, “which describes the usability of the information from the time of creation to publication and finally archiving”(Sandrini 2008, 10).

The information is stored in web documents, using markup languages such as XML (eXtensible Markup Language) or HTML (HyperText Markup Language) (ibid.). In the following section, we define the latter to have a clear understanding of the markup language we will work with.

2.1.2 HyperText Markup Language

The acronym HTML stands for HyperText Markup Language and refers to the core language of the World Wide Web (W3C 2018). In 1989, when Tim Berners-Lee invented the Web, he used HTML as its publishing language with the idea of gathering information from researchers all around the world and link all their studies to each other, to obtain cross-references from one research to another. In simpler words, HTML is the text format of websites (Raggett et al.

1998). In its first ten years, until 1998, this markup language kept developing, resulting in the HTML 4.0 version. As people in the industry started focusing on different types of languages to provide web content, HTML was put aside. However, as new technologies such as XForms were introduced, the industry started working on a new version, introduced in 2014: HTML 5.0 (W3C 2018). The latest version, HTML 5.3, was published on the 18^th October 2018³.

HTML5 is based on four design principles: compatibility, utility, interoperability, and universal design. The latter concerns directly our research, as it includes accessibility among its three concepts (the other two being media independence and support for all world languages).

According to this concept, design features should be accessible to everyone regardless of their abilities, particularly to people with disabilities (Lawson & Sharp 2010, 3-5; Lubbers, Albers

& Salim 2010, xiii; W3C 2007).

Moreover, HTML5 specifications include what has previously been specified in HTML 4.1, XHTML⁴ 1.1 and DOM⁵ Level 2 HTML (W3C, 2018) and “it has moved HTML from being a relatively simple document markup language to being a sophisticated platform for web applications with a host of new, rich application programming interfaces (APIs)” (O Connor 2012, 1). Therefore, new features and new elements (see Appendix A) were introduced in this

3 https://www.w3.org/TR/html53/introduction.html#introduction [15.6.2019]

4 Extensible HyperText Markup Language: an almost identical but stricter HTML defined as an XML (eXtensible Markup Language) application.

5 Document Object Model: it defines HTML elements as objects and their properties, methods, and events. It is also an API for JavaScript.

(29)

latest version, which often presents many challenges to users, and especially to people with disabilities (ibid.).

This new version has a host of elements that include a broad range of new functions. First, developers introduced new semantics: elements, attributes, and attribute values have specific meanings and they are used only for their intended purpose (O Connor 2012, 128; W3C 2018a).

Thus, new semantic markup is used to describe the element’s content (Lubbers et al. 2010, 10).

In the context of our research, we deem it important to emphasise that semantic information is essential to ensure accessibility, as assistive technologies use it to provide information to present the document and to provide additional functionalities (W3C 2018a).

We can use semantic information to structure the document. HTML5 provides many new sectioning elements, such as <header> (which includes the header’s content, at the top of the page) and <section> (which include a section’s content in the document’s body), as we illustrated in Figure 2.1.

Figure 2.1: Basic structure of an HTML document (adapted from O Connor 2012, 167; Lubbers et al. 2010, 11-12).

Second, every element includes information about the category it belongs to, the context it is used in, the ‘content model’ that defines the element’s content, and the DOM interface implemented (O Connor 2012, 142-3). Concerning the category the element belongs to, we can find different types of content; however, one element can have different characteristics, and for this reason, it can fall under several categories. As illustrated in Table 2.1, content can be divided into the following categories: metadata, flow, sectioning, heading, phrasing, embedded, and interactive (Lubbers et al. 2010, 10; O Connor 2012, 143-52; W3C 2018a).

(30)

Content model Description

Metadata content This kind of content outlines the presentation of the page, the behaviour of the content, and the relationships with other documents (W3C 2018a).

Flow content It includes all the main elements used in the body of a document (ibid.).

Sectioning content It defines new sections that can be grouped together and defines the scope of a certain part of the document, such as headings and footers (O Connor 2012, 147; W3C 2018a).

Heading content This kind of content helps determine the structure of the page and defines the header of a section (O Connor 2012, 147; W3C 2018a).

Phrasing content “This is the main body of text in a document and the inline elements used to mark up the content” (O Connor 2012, 147). In the context of content models, the term ‘text’ defines nothing or text nodes, which consist of Unicode characters⁶ (W3C 2018a).

Embedded content This kind of content allows the import of another source in the HTML document (O Connor 2012, 150).

Interactive content It defines the “content that is specifically intended for user interaction” (W3C 2018a).

Table 2.1: Content model categories.

Last, among all the new HTML5’s elements, we should talk about one crucial aspect in the context of accessibility and our research: WAI-ARIA. WAI-ARIA (Accessible Rich Internet Applications) is a “technical specification that provides a framework to improve the accessibility and interoperability of web content and applications” (W3C/WAI 2017b). This specification is essential to ensure accessibility, as it allows to include more information than the current HTML5 specification can provide (O Connor 2012, 135). An ARIA element is usually added to the native HTML element (ibid., 140) and, except cases where there are restrictions, web developers can use the ARIA role or aria-* attributes (W3C 2018a). For instance, when an HTML element is followed by the attribute hidden, the author can

6 The Unicode Standard ist he universal character encoding standard for written characters and text:

https://www.unicode.org/versions/Unicode12.0.0/ch01.pdf [17.6.2019]

(31)

introduce the ARIA semantics ‘aria-hidden’ and set it to ‘true’ (O Connor 2012, 139). In this way, a screen reader will ignore the element and avoid to read it to the user.

2.2 Web accessibility

As we have seen in Chapter 1, web accessibility consists of developing websites and technologies that are designed for everyone, in particular people with disabilities. In the following section, we briefly examine the WCAG guidelines, as they provide the basis for the selection of the research material and the selection of a set of requirements.

2.2.1 Web Content Accessibility Guidelines

The Web Content Accessibility Guidelines (WCAG), published for the first time in 1999, are a set of recommendations for making the content of the Web accessible. It was developed through the cooperation of W3C (specifically of the Accessibility Guidelines Working Group), individuals, and organisations around the world. The latest version of this document, WCAG 2.1 (Kirkpatrick Andrew et al. 2018), was published on the 5^th June 2018. The set of guidelines is divided into four principles (perceivable, operable, understandable, and robust); thirteen guidelines, which provide basic goals to authors in order to make accessible content; and 77 success criteria (SC), divided into three levels of conformance (A, AA, AAA) and written as testable statements that are not technology-specific, and followed by a wide variety of techniques (ibid.), 597 in total.

Techniques offer specific guidance for web developers and evaluators to meet the WCAG success criteria and make their websites accessible and are divided into three categories:

sufficient techniques, advisory techniques, and failures. The first type includes the techniques that should be implemented, while the second is “suggested to improve accessibility”

(W3C/WAI 2019b), but they may not be sufficient and web browsers and/or assistive technologies may not support them. On the contrary, failures define the cause of accessibility barriers and, therefore, they should be avoided (ibid.).

Moreover, techniques can also be general or technology-specific. In the following table (Table 2.2), we briefly describe the different types of techniques.

(32)

Techniques Description

General In this category, we find, as the name says, general techniques. It is important to clarify that technology-specific techniques do not replace the general ones, but they should be considered at the same level (W3C/WAI 2019c).

ARIA As we have seen in section 2.1.2, WAI-ARIA helps to add information in HTML elements. ARIA techniques provide ways to implement these elements and make them accessible, with the purpose of describing the content of the web page (ibid.).

Client-Side Script This kind of techniques includes all the scripts that the client, or the user, can see. In this way, the user can modify and manipulate the content of the page according to their needs (ibid.).

CSS CSS (Cascading Style Sheet) is used to modify the appearance of a web page and can be used to resolve issues related to contrasts, font size, and media support (ibid.).

Flash This type of technique concerns Adobe Flash Player (a cross-platform browser plug-in), which allows adding videos, vector-based graphics, and other elements to the website. Therefore, it is crucial for authors to follow these techniques so that this kind of content is still accessible to people with disabilities.

HTML These techniques give recommendations concerning HTML and XHTML, namely how to ensure accessibility by using the appropriate HTML elements and attributes (ibid.).

PDF PDF (Portable Document Format) is a file-format that represents independent documents. This type of techniques helps authors describe the logical order of content and present the content of each part (ibid.) Server-Side Script Contrary to Client-Side Script, this type of technique includes all the scripts that the server itself should manipulate to ensure accessibility (ibid.).

Silverlight Microsoft Silverlight is a tool used for creating interactive user experiences and mobile applications. Built-in support is provided by Silverlight for accessibility (ibid.).

(33)

SMIL SMIL (Synchronized Multimedia Integration Language) is a markup language used for multimedia elements, such as videos and audios. It is important for the author to add descriptions and captions when dealing with this type of content (ibid.).

Plain-Text This technique concerns the text format for paragraphs, lists, and headings (ibid.).

Table 2.2: WCAG techniques.

However, to ensure accessibility, we also need to consider several components of web development, namely content, web browsers, media players, assistive technology, users, developers, authoring tools, and evaluation tools (W3C/WAI 2018a). To address all these aspects, WAI published two additional documents, which together with the WCAG constitute the three-part approach of W3C: the Authoring Tool Accessibility Guidelines (ATAG)⁷ and the User Agent Accessibility Guidelines (UAAG)⁸.

2.3 Web accessibility and web localisation

In Section 1.1.1 and Section 2.1.1, we defined both web accessibility and web localisation, respectively. But how are these two concepts related to each other? As we introduced in the first chapter, localisers can contribute to the achievement of web accessibility (Rodríguez Vázquez 2013). In this section, we will analyse some articles and studies by different scholars, who examined the relationship between web accessibility and localisation.

In an informational article, Ó Broin (2004) illustrates how he initially believed that there might be a conflict between accessibility requirements and localisation. However, he then realised that localisation could be, in fact, a form of accessibility, because through localisation we deliver equivalent information and, consequently, make the content accessible to different audiences. In addition, he also listed several areas that concern both localisation and accessibility: language detection, a clear understanding of texts, text externalisation from graphics and multimedia, abbreviations and acronyms, tabular data, and separation of content from presentation. As we will see in future sections, these elements appear often on accessible websites and, according to the author, localisers should be aware of these specific accessibility features, as they will have positive implications on the general localisation process (ibid.).

7 https://www.w3.org/WAI/standards-guidelines/atag/ [11.6.2019]

8 https://www.w3.org/WAI/standards-guidelines/uaag/ [11.6.2019]

(34)

In another educational article titled “Fostering accessibility through localisation”, Rodríguez Vázquez and Torres Del Rey (2014) embrace this idea of localisation as a form of accessibility and highlight how “accessibility achievements in the monolingual site could be undone” (ibid., 35) if localisers do not transfer the same level of accessibility achieved in the original code.

The authors state that the final target product should be as accessible as the original one and that localisers could contribute to web accessibility audits by working with the original development team of the product, sharing knowledge and collaborating among the actors of the localisation workflow (ibid., 36-37).

In a subsequent article focusing on accessibility as a key factor in localisers education, the same authors (2016) argue that “[t]raining localisers from the (disad)vantage point of accessibility allows teachers and students to become aware of what the product does, means, can be used for; who, and how, its beneficiaries and users would be; and, most importantly, how the above is achieved and can be achieved (or adapted for functional diversity) in different locales or cultural contexts” (Torres del Rey & Rodríguez Vázquez 2016, 975). Consequently, when a localiser is aware of how accessibility best practices are implemented, the degree of accessibility achieved in the target product could be superior compared to the source. This result was demonstrated in a study carried out by Rodríguez Vázquez (2016), who focused on images localisation: accessibility can be achieved not only through the translation of text alternatives, but also through the assessment of their suitability (ibid.). We consider this aspect significant for our research, as we will also try to determine whether the level of localisers knowledge of accessibility can have an impact on the final degree of accessibility.

Nevertheless, from an industry perspective, a qualitative study carried out by Rodríguez Vázquez and O’Brien (2017), in which fifteen representatives of six language service providers were interviewed, suggested that there is still a lack of awareness on the importance of accessibility and, consequently, it is not automatically taken into account during the localisation process (ibid.).

2.3.1 Multilingual web accessibility

The element of multilingualism is barely contemplated in studies and official documents that concern accessibility. In 2016, Rodríguez Vázquez argued that “at present there is no standardised procedure for assessing the accessibility level of multilingual website” (Rodríguez Vázquez 2016, viii).

(35)

Since then, this aspect has been further studied in several theses. Casalegno (2018) investigated the impact of partial localisation strategies on the web navigation experience of screen reader users when browsing multilingual websites. Her study demonstrated that users encountered more difficulties in navigating localised versions of a multilingual website compared to the original version, due to language-related issues, technical issues (which concern the interaction between websites and screen readers), and lack of clarity in the website structure. In conclusion, she could determine that users encountered fewer usability issues in the original language version than in the localised versions (ibid.).

One year later, Pontus (2019) evaluated the degree of accessibility of a sample of 50 airline websites by performing both automated and manual checks. She noticed several general accessibility issues, mainly related to online forms, ticket purchasing, and the reservation system. In addition, she encountered more accessibility problems in the French and German (localised) versions than in the original English versions, as English “values tend to ‘travel’ to localised sites in other languages and create language-related accessibility problems”(ibid., 112). Overall, these two recent investigations have shown that accessibility continues to be an issue in the multilingual web and hence motivates the pertinence of yet another study in this domain.

2.3.2 Accessibility features

Torres del Rey and Morado Vázquez (2019) contributed to the research currently being conducted on localisation and accessibility by asking the following question: “does accessibility have a concrete form or clearly defining characteristics, and can the forms and characteristics that are culture- and language-bound be “captured” formally?” (ibid.).

Accessibility can be seen as a ‘quality’ (Jiménez-Crespo 2013, 126-31; Rodríguez Vázquez 2016, 62-4; Torres del Rey & Morado Vázquez 2019), and as such, it cannot be transferred.

However, as we saw in Section 2.2.1, there are a series of recommendations that can be implemented to achieve the same or even a better level of accessibility. As a result, we could state that localisers could indeed transfer these “characteristics”, embedded in the form of coding elements, attributes, text units, and relations that help achieve this qualit y (Torres del Rey & Morado Vázquez 2019). In the context of this thesis, this is what we call ‘accessibility features’.

The authors continue by making a distinction between ‘neutrally transferable’ (embedded in the code’s structure) and ‘re-placed’ features (in-line formatting), which are usually excluded

Localisation and web accessibility: a comparative study of the transfer of accessibility information through CAT tools

Master

Reference

Localisation and web accessibility: a comparative study of the transfer of accessibility information through CAT tools

Isotta Pacati

Localisation and web accessibility: a comparative study of the transfer of

accessibility information through CAT tools

Directrice : Silvia Rodriguez Va ́zquez Jurée : Luci ́a Morado Vázquez

Déclaration attestant le caractère original du travail effectue ́

Acknowledgments

Abstract

Contents

List of Figures

List of Tables

List of Abbreviations

1 Introduction

1.1 Research context

1.2 Motivation

1.3 Research questions and hypotheses

1.4 Methods

1.5 Structure of the thesis

2 Literature Review

2.1 Localisation

2.2 Web accessibility

2.3 Web accessibility and web localisation